RCLSMIX-class {rebmix}

Class "RCLSMIX"

Description

Object of class RCLSMIX.

Objects from the Class

Objects can be created by calls of the form new("RCLSMIX", ...). Accessor methods for the slots are a.o(x = NULL), a.Dataset(x = NULL), a.s(x = NULL), a.ntrain(x = NULL), a.P(x = NULL), a.ntest(x = NULL), a.Zt(x = NULL), a.Zp(x = NULL), a.CM(x = NULL), a.Accuracy(x = NULL), a.Error(x = NULL), a.Precision(x = NULL), a.Sensitivity(x = NULL), a.Specificity(x = NULL) and a.Chunks(x = NULL), where x stands for an object of class RCLSMIX.

Slots

x:

a list of objects of class REBMIX of length o obtained by running REBMIX on g = 1, \ldots, s train datasets Y_{\mathrm{train}g} all of length n_{\mathrm{train}g}. For the train datasets the corresponding class membership \bm{\Omega}_{g} is known. This yields n_{\mathrm{train}} = \sum_{g = 1}^{s} n_{\mathrm{train}g}, while Y_{\mathrm{train}q} \cap Y_{\mathrm{train}g} = \emptyset for all q \neq g. Each object in the list corresponds to one chunk, e.g., (y_{1j}, y_{3j})^{\top}.

o:

number of chunks o. Y = \{\bm{y}_{j}; \ j = 1, \ldots, n\} is an observed d-dimensional dataset of size n of vector observations \bm{y}_{j} = (y_{1j}, \ldots, y_{dj})^{\top} and is partitioned into train and test datasets. Vector observations \bm{y}_{j} may further be split into o chunks when running REBMIX, e.g., for d = 6 and o = 3 the set of chunks substituting \bm{y}_{j} may be as follows (y_{1j}, y_{3j})^{\top}, (y_{2j}, y_{4j}, y_{6j})^{\top} and y_{5j}.

Dataset:

a data frame containing test dataset Y_{\mathrm{test}} of length n_{\mathrm{test}}. For the test dataset the corresponding class membership \bm{\Omega}_{g} is not known.

s:

finite set of size s of classes \bm{\Omega} = \{\bm{\Omega}_{g}; \ g = 1, \ldots, s\}.

ntrain:

a vector of length s containing numbers of observations in train datasets Y_{\mathrm{train}g}.

P:

a vector of length s containing prior probabilities P(\bm{\Omega}_{g}) = \frac{n_{\mathrm{train}g}}{n_{\mathrm{train}}}.

ntest:

number of observations in test dataset Y_{\mathrm{test}}.

Zt:

a factor of true class membership \bm{\Omega}_{g} for the test dataset.

Zp:

a factor of predictive class membership \bm{\Omega}_{g} for the test dataset.

CM:

a table containing confusion matrix for multiclass classifier. It contains number x_{qg} of test observations with the true class q that are classified into the class g, where q, g = 1, \ldots, s.

Accuracy:

proportion of all test observations that are classified correctly. \mathrm{Accuracy} = \frac{\sum_{g = 1}^{s} x_{gg}}{n_{\mathrm{test}}}.

Error:

proportion of all test observations that are classified wrongly. \mathrm{Error} = 1 - \mathrm{Accuracy}.

Precision:

a vector containing proportions of predictive observations in class g that are classified correctly into class g. \mathrm{Precision}(g) = \frac{x_{gg}}{\sum_{q = 1}^{s} x_{qg}}.

Sensitivity:

a vector containing proportions of test observations in class g that are classified correctly into class g. \mathrm{Sensitivity}(g) = \frac{x_{gg}}{\sum_{q = 1}^{s} x_{gq}}.

Specificity:

a vector containing proportions of test observations that are not in class g and are classified into the non g class. \mathrm{Specificity}(g) = \frac{n_{\mathrm{test}} - \sum_{q = 1}^{s} x_{qg}}{n_{\mathrm{test}} - \sum_{q = 1}^{s} x_{gq}}.

Chunks:

a vector containing selected chunks.

Author(s)

Marko Nagode

References

D. M. Dziuda. Data Mining for Genomics and Proteomics: Analysis of Gene and Protein Expression Data. John Wiley & Sons, New York, 2010.